The memorial's printed record, set against a separate archive
First Appears upon the Record
Henry Krehbiel's 1892 memorial appendix prints the 259 regular concerts it records from 1842 to April 1892, not every known Society event, with marks printed in the book such as first time, first time in America, and new. Transcribed from the page images, its 259 concerts and 214 printed marks are set concert by concert against the New York Philharmonic's open performance archive. The marks cannot choose between counting whole works and counting movements, since a rule that calls everything new agrees with every one; the disagreements, a recurrence test the marks themselves supply, and 183 seasons of subscription concerts show what each choice does.
Nine of the 214 printed marks in the Philharmonic Society's 1892 memorial are contradicted by the open Performance History catalogue, counting whole works: six parts or versions filed under one work, two concerts the memorial omits, one self-contradiction.
The title is Krehbiel's phrase (page 89): Schubert "first appears upon the record January 11, 1851." The memorial transcription and the Performance History metadata are separate catalogues with overlapping concerts. This page studies where their lists and first-appearance readings disagree.
Layer one | put a hand on the record
Pick a concert. Read the memorial's marks beside the archive.
Left, the programme as Krehbiel printed it, with labels printed in the memorial. Right, the archive's rows for the paired concert. Each printed item carries the archive's verdict under the counting rule you choose. Start with the First Concert of November 19, 1859, then switch the unit.
Archive rows
An item without a mark is never read as a repeat. The memorial did not mark every possible first appearance in this archive, so a missing mark is not evidence.
Every mark, every free choice
Where the marks and the archive disagree
Every engine choice below changes a relevant output. The defaults count whole works, count the Philharmonic's own concerts of every kind as its memory, and pair a printed concert with the nearest archive concert within three days.
Only active for movement counts.
| Printed concert | Printed item and mark | Archive's earlier row | Computed tags | Note |
|---|
Layer two | answer the dismissal
"A movement is part of its work. This is pedantry."
The marks cannot settle the unit on their own: they only ever say new. But the printed record supplies a test that can fail in both directions, and unit, memory, event scope and smoothing choices go on moving a novelty curve long after the memorial stops.
A test the marks themselves supply
When the appendix marks a piece and later prints the same piece again, unmarked, the printed record gives us a repeat to test. Any counting rule worth using must call it one. The pairs were found from the printed words and read by hand; parts, generic titles and different pieces are held out and counted.
| Rule | Repeats | Called new | Which later performances it calls new |
|---|
183 seasons of subscription concerts
The share of each season's distinct pieces that the archive has never recorded before, the first-appearance-in-this-record measure Eric William Lin charted in 2017. The chart follows unit, memory, event scope and smoothing. Mark kind, concert pairing and multi-row reading affect the mark and recurrence checks, not this season series. Drag the slider or move across the chart to read a season.
Lin wrote that repeats outnumbered first-time pieces "by the 1859-60 season", counting subscription concerts only and leaving out non-orchestral pieces and encores, which this page does not do. His figures are a comparison, not an anchor.
Every season as a table
The other direction: one piece, several IDs
The marks cannot see this. The memorial prints Rubinstein's "Ocean" Symphony again and again under one name; the archive files three versions under different base works, so counting whole works calls a repeat new.
It happens after 1892 too, with no witness to say which reading is right: the Tristan Act III prelude is a movement of the opera and, separately, an arrangement catalogued as its own work.
Whose memory is the merged orchestra's?
The archive holds the New York Symphony's concerts beside the Philharmonic's. The memorial prints Brahms's Fourth Symphony with the label "(new)" at the Philharmonic on January 15, 1887.
Whether a merged orchestra inherits another's memory is a choice. No answer to it is in either record.
The check | shown inside the work
What the page recomputes, and what it cannot
Anchor: Krehbiel's own date
His two hundred and fifty-nine concerts
"the programmes of the two hundred and fifty-nine regular concerts which it has given since its organization." (pages 8 and 9)
Readings of the 259, compared as lists, not totals
List agreement compares each engine pairing with the frozen archive descriptors in the shipped match table. Total agreement is a separate count.
| Reading | Total | Total agrees | List agrees | Why not |
|---|
Archive concerts the memorial does not print
After the memorial ends
Printed and archive dates that differ
Both dates are shown; the page never picks one. The printed and archive weekdays do not settle which date is right.
| Heading as printed | Printed | Archive |
|---|
Krehbiel's other first-appearance dates, pages 87 to 92
Every dated sentence, printed against the engine
| Page | As printed | Printed | Engine | Agrees |
|---|
The transcription
Every concert heading, item and mark on printed pages 95 to 163 was typed from the page images of the Internet Archive scan, then compared word by word with the scan's OCR. Where they differed the image was read again. The counts below are the engine's, from that transcription.
Free choices, currently
Uncertainty kept in view
- The match from printed item to archive row is a judgment: first proposed by a matcher, then read by hand for every marked item and every leftover. The reasons for every hand decision ship with the page.
- Whose marks these are is not settled. The vocabulary changes by period, which suggests they follow the original printed programmes, but no original programme was checked. The page says "marks as printed in Krehbiel 1892".
- The archive can contradict a "(first time in America)" mark with an earlier row of its own. It can never confirm one.
- The memorial's narrative dates the Lohengrin prelude November 19, 1857; its own appendix and the archive put it on November 19, 1859. The page reports the disagreement and does not correct the book.
The edge of the instrument
Ask it for something the records cannot carry
Choose a question to see why the page will not answer it.
Apparatus and prior art
The sources stay in view
The witness
Henry Edward Krehbiel, The Philharmonic Society of New York: A Memorial (New York and London: Novello, Ewer and Co., 1892), Appendix, "Programmes of the Concerts of the Philharmonic Society", pages 95 to 163. Public domain. Scan: Internet Archive, cu31924022356822 (Cornell University Library).
The archive
New York Philharmonic Performance History, nyphilarchive/PerformanceHistory at commit ee2d2b40, metadata released under CC0 1.0. With thanks and attribution to the New York Philharmonic, which neither made nor endorses this page. The metadata here is modified: flattened, reduced to the fields the page uses, intervals dropped, dates cut to the day. It is a separate catalogue from Krehbiel's memorial; this comparison uses their overlapping concerts. The Philharmonic's Data Explorer and API are rights reserved and were not used.
Prior art
Eric William Lin, "Pulsecheck" (2017), charted first-time and living-composer shares from this archive. Jake Gibson's programming analysis (2024) counts new works per season. Kris Shaffer (2017) flattened the data with movement-aware counts; Andrew Gustar (2019) studied programme IDs against dated concerts. John Erskine's 1943 history continues the institutional record with later subscription-programme listings, but does not reconcile Krehbiel's marks with this archive. Dowd, Liddle, Lupo and Borden, Poetics 30 (2002), and Kremp, Social Forces 88.3 (2010), studied American orchestral repertoires.
Stevenson, "Gottschalk Programs Wagner" (1983) gives the Philharmonic's Lohengrin prelude as 19 November 1859. That establishes the 1859 date used in the page's date correction, but it does not identify Krehbiel's internal contradiction or compare the narrative and appendix with the archive.
Music in Gotham's entry for the 1873 concert reproduces a review saying that the Philharmonic introduced Rubinstein's Ocean Symphony in four movements and that the 1873 performance had six after later additions. That establishes the historical version split used by the page, but it does not use Performance History IDs, audit a Krehbiel mark, or establish the page's recurrence result.
We searched the New York Philharmonic Performance History repository, public analyses of its data, bibliographic records for Erskine's history, the repertoire studies of Dowd et al. 2002 and Kremp 2010, and Krehbiel's 1892 memorial on 2026-09-15 and did not find a published work that anticipates the full reconciliation of the 214 marks and 259 concerts against the archive. Two individual examples have antecedents, now credited above. This bounded search is not a claim of exhaustive absence.